-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathld_score_comparison_figures.py
555 lines (405 loc) · 19.3 KB
/
ld_score_comparison_figures.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
from scipy import stats
import numpy as np
import pandas as pd
import matplotlib.pylab as plt
import seaborn as sns
from utils import makedir, write_pbz2
from collections.abc import Iterable
from multiprocessing import Pool
import os
def evaluate_model_differences(joint_df):
lds_pairs = [
['R2_0.0', 'D2_0.0'],
['R2_1.0', 'D2_1.0'],
['D2_1.0', 'D2_0.0'],
['D2_1.0', 'D2_0.025'],
['D2_1.0', 'D2_0.5'],
['D2_1.0', 'D2_0.75']
]
results = {}
for sc1, sc2 in lds_pairs:
try:
reg_df = joint_df[['SNP', 'CHR', 'MAF', sc1, sc2]]
except Exception:
continue
slope, intercept, r_value, p_value, std_err = stats.linregress(reg_df[sc1],
reg_df[sc2])
results[f"{sc2} ~ {sc1}"] = {
'slope': slope,
'intercept': intercept,
'R-squared': r_value**2,
'p-value': p_value
}
reg_df['Diff'] = reg_df[sc2] - reg_df[sc1]
slope, intercept, r_value, p_value, std_err = stats.linregress(reg_df['MAF'],
reg_df['Diff'])
results[f"{sc2} - {sc1} ~ MAF"] = {
'slope': slope,
'intercept': intercept,
'R-squared': r_value ** 2,
'p-value': p_value
}
return results
def plot_manhattan_ld(joint_df, ld_col, title,
output_fname, pos_col='BP', alpha=0.3):
starting_pos = 0
ticks = []
chrom_spacing = 25000000
plt.figure(figsize=(18, 9))
plt.axhline(0.0, ls='--', zorder=1,
color='#263640')
plt.axhline(joint_df[ld_col].mean(), ls='--', zorder=1,
color='#D60A1F', label='Mean')
unique_chr = sorted(joint_df['CHR'].unique())
for i, ch in enumerate(unique_chr):
chr_df = joint_df.loc[joint_df['CHR'] == ch]
max_pos = chr_df[pos_col].max()
xmin = starting_pos - (chrom_spacing / 2)
xmax = max_pos + starting_pos + (chrom_spacing / 2)
if i % 2 == 1:
plt.axvspan(xmin=xmin, xmax=xmax, zorder=0, color='#F8F8F8')
ticks.append((xmin + xmax) / 2)
plt.scatter(chr_df[pos_col] + starting_pos, chr_df[ld_col],
c=chr_df['MAF'], cmap='cool', alpha=alpha, label=None,
marker=scatter_mrkr, s=2.0)
starting_pos += max_pos + chrom_spacing
plt.xlim([-chrom_spacing, starting_pos])
plt.xticks(ticks, unique_chr)
cbar = plt.colorbar(pad=0.01)
cbar.set_label('MAF')
if 'RELDIFF' in ld_col:
plt.ylim(-0., 100.)
plt.xlabel("Genomic Position")
plt.ylabel(ld_score_names[ld_col])
if include_title:
plt.title(title)
plt.legend()
plt.tight_layout()
plt.savefig(output_fname)
plt.close()
def plot_ld_score_comparison_scatter(joint_df,
compare_against,
title,
output_fname,
base_score="R2_1.0",
chr_num=None,
color_col=None):
if not isinstance(compare_against, Iterable) or type(compare_against) == str:
compare_against = [compare_against]
if chr_num is not None:
joint_df = joint_df.loc[joint_df['CHR'] == chr_num, ]
plt.figure(figsize=figure_size)
for sc in compare_against:
if len(compare_against) > 1:
label = ld_score_names[sc]
else:
label = None
if color_col is None:
plt.scatter(joint_df[base_score], joint_df[sc],
color=['skyblue', None][len(compare_against) > 1],
marker=scatter_mrkr, alpha=0.3,
label=label)
else:
plt.scatter(joint_df[base_score], joint_df[sc],
c=joint_df[color_col], cmap='cool',
marker=scatter_mrkr, alpha=0.3,
label=label)
if 'RELDIFF' in compare_against[0]:
plt.ylim(-100., 100.)
else:
# Plots line
x = np.linspace(0.0, joint_df[compare_against + [base_score]].max().max(), 1000)
plt.plot(x, x, linestyle='--', color='#D60A1F')
plt.xlabel(ld_score_names[base_score])
if len(compare_against) > 1:
plt.ylabel("Unbiased LD Score")
else:
plt.ylabel(ld_score_names[compare_against[0]])
if color_col is not None:
cbar = plt.colorbar(pad=0.05)
cbar.set_label(color_col)
if len(compare_against) > 1:
plt.legend()
if include_title:
plt.title(title)
plt.tight_layout()
plt.savefig(output_fname)
plt.close()
def plot_cross_population_comparison(pop1_df, pop1_name,
pop2_df, pop2_name,
ld_col, output_fname):
joint_df = pd.merge(pop1_df[['SNP', ld_col]],
pop2_df[['SNP', ld_col]],
on='SNP',
suffixes=('_' + pop1_name, '_' + pop2_name))
plt.figure(figsize=figure_size)
sns.scatterplot(x=ld_col + '_' + pop1_name,
y=ld_col + '_' + pop2_name,
marker=scatter_mrkr,
data=joint_df, alpha=0.3, color='skyblue')
if 'DIFF' in ld_col:
plt.axhline(y=0.0, color='grey', linestyle='--')
plt.axvline(x=0.0, color='grey', linestyle='--')
else:
x = np.linspace(0.0, joint_df[[ld_col + '_' + pop1_name,
ld_col + '_' + pop2_name]].max().max(), 1000)
plt.plot(x, x, linestyle='--', color='grey')
plt.errorbar(joint_df[ld_col + '_' + pop1_name].mean(),
joint_df[ld_col + '_' + pop2_name].mean(),
xerr=joint_df[ld_col + '_' + pop1_name].std(),
yerr=joint_df[ld_col + '_' + pop2_name].std(),
ecolor='#D60A1F', capsize=2)
plt.xlabel(pop1_name)
plt.ylabel(pop2_name)
if include_title:
plt.title(ld_score_names[ld_col])
plt.tight_layout()
plt.savefig(output_fname)
plt.close()
def plot_sample_size_comparison(ref_pop='EUR',
sample_pop=('EUR20', 'EUR50', 'EUR100', 'EUR200'),
comp_ld_scores=('R2_1.0', 'D2_1.0', 'UR2_1.0')):
res = []
comp_ld_scores = list(comp_ld_scores)
for sp in sample_pop:
joint_df = pd.merge(pop_scores[ref_pop][['SNP'] + comp_ld_scores],
pop_scores[sp][['SNP'] + comp_ld_scores],
on='SNP',
suffixes=('_' + ref_pop, '_' + sp))
for lds in comp_ld_scores:
slope, intercept, r_value, p_value, std_err = stats.linregress(joint_df[f"{lds}_{sp}"],
joint_df[f"{lds}_{ref_pop}"])
res.append({
'Sample Size': int(sp.replace(ref_pop, "")),
'Estimator': lds,
'R^2': r_value**2,
'Relative bias': np.mean((joint_df[f"{lds}_{sp}"] -
joint_df[f"{lds}_{ref_pop}"]) / joint_df[f"{lds}_{ref_pop}"])
})
res_data = pd.DataFrame(res)
res_data.to_csv(os.path.join(analysis_dir, "sample_size_comparison.csv"))
estimator_labels = ['Corrected $r^2$', 'Unbiased $D^2$', 'Raw $r^2$']
plt.figure(figsize=figure_size)
g = sns.lineplot(x="Sample Size", y="R^2",
hue="Estimator", style="Estimator",
palette=ld_scores_colors, markers=True,
dashes=False, data=res_data)
for t, l in zip(g.legend().texts[1:], estimator_labels):
t.set_text(l)
plt.tight_layout()
plt.savefig(os.path.join(plot_dir, "sample_size_comparison_r2.png"))
plt.close()
plt.figure(figsize=figure_size)
g = sns.lineplot(x="Sample Size", y="Relative bias",
hue="Estimator", style="Estimator",
palette=ld_scores_colors, markers=True,
dashes=False, data=res_data)
for t, l in zip(g.legend().texts[1:], estimator_labels):
t.set_text(l)
plt.tight_layout()
plt.savefig(os.path.join(plot_dir, "sample_size_comparison_relative_bias.png"))
plt.close()
def get_univariate_predictor(lds_df, maf_df, scaled=True):
"""
The univariate predictor is the term in the univariate
LD Score regression that doesn't depend on h_g and N.
e.g. in the Bulik-Sullivan et al. equation, uni_pred = (lj/M).
For the purposes of numerical stability, it may be desirable
to work with the scaled univariate predictor M*uni_pred.
"""
for ld_col in lds_df.columns:
if ld_col in ld_scores:
alpha = float(ld_col.split("_")[-1])
lds_df[ld_col] /= np.mean(maf_df['VAR']**(1. - alpha))
if not scaled:
for ldn in ld_scores:
lds_df[ldn] /= len(maf_df)
else:
return lds_df
def read_merge_ld_scores(chr_num):
ld_dfs = []
processed_ld_scores = []
for i, lds in enumerate(ld_scores):
try:
if i == 0:
ld_dfs.append(pd.read_csv(modified_ld_score_dir % (pop, lds, chr_num),
sep="\t", index_col=1)[cols_to_keep + ['base' + lds]])
ld_dfs[-1]['SNP'] = ld_dfs[-1].index
else:
ld_dfs.append(pd.read_csv(modified_ld_score_dir % (pop, lds, chr_num),
sep="\t", index_col=1)[['base' + lds]])
processed_ld_scores.append(lds)
except Exception as e:
print(e)
continue
chr_df = pd.concat(ld_dfs, axis=1)
frq_df = pd.read_csv(frq_file % (pop, pop, chr_num), sep="\s+")
frq_df['VAR'] = 2*frq_df['MAF']*(1. - frq_df['MAF'])
chr_df.columns = cols_to_keep + processed_ld_scores[:1] + ['SNP'] + processed_ld_scores[1:]
return chr_df, frq_df
if __name__ == '__main__':
plot_dir = "figures/ld_score_comparison_test/"
analysis_dir = "results2/analysis/ld_score_comparison_test"
num_procs = 6
figure_size = (7.5, 6)
scatter_mrkr = '.'
fig_format = '.png'
sns.set_context('talk')
include_title = False
populations = ['EUR', 'EUR20', 'EUR50', 'EUR100', 'EUR200', 'EAS', 'AFR']
ld_scores = ['R2_0.0', 'R2_0.25', 'R2_0.5', 'R2_0.75', 'R2_1.0',
'D2_0.0', 'D2_0.25', 'D2_0.5', 'D2_0.75', 'D2_1.0',
'UR2_0.0', 'UR2_1.0']
ld_score_diff = ['DIFF_D2_1.0_R2_1.0'] #['DIFF_LD2_L2', 'DIFF_LD2MAF_L2MAF', 'DIFF_LD2_LD2MAF']
rel_ld_score_diff = ['RELDIFF_D2_1.0_R2_1.0'] # ['RELDIFF_LD2_L2', 'RELDIFF_LD2MAF_L2MAF', 'RELDIFF_LD2_LD2MAF']
ld_scores_colors = {
'D2_0.0': '#E15759',
'D2_1.0': '#76B7B2',
'R2_0.0': '#B07AA1',
'R2_1.0': '#F28E2B',
'UR2_0.0': '#1f77b4',
'UR2_1.0': '#1f77b4'
}
ld_score_names = {
'R2_0.0': '$r^2 (\\alpha=0)$ LD Score',
'R2_0.25': '$r^2 (\\alpha=0.25)$ LD Score',
'R2_0.5': '$r^2 (\\alpha=0.5)$ LD Score',
'R2_0.75': '$r^2 (\\alpha=0.75)$ LD Score',
'R2_1.0': '$r^2 (\\alpha=1)$ LD Score',
'D2_0.0': '$D^2 (\\alpha=0)$ LD Score',
'D2_0.25': '$D^2 (\\alpha=0.25)$ LD Score',
'D2_0.5': '$D^2 (\\alpha=0.5)$ LD Score',
'D2_0.75': '$D^2 (\\alpha=0.75)$ LD Score',
'D2_1.0': '$D^2 (\\alpha = 1)$ LD Score',
'DIFF_LD2_L2': 'LD Score Difference $D^2 - r^2$ ($\\alpha = 0$)',
'DIFF_LD2MAF_L2MAF': 'LD Score Difference $D^2 - r^2$ ($\\alpha = 1$)',
'DIFF_LD2_LD2MAF': 'LD Score Difference $D^2 (\\alpha = 0) - D^2 (\\alpha = 1)$',
'RELDIFF_LD2_L2': '% Difference in LD Scores ($\\alpha = 0$)\n$abs(D^2 - r^2) / r^2$',
'RELDIFF_LD2MAF_L2MAF': '% Absolute Difference in LD Scores ($\\alpha = 1$)\n$abs(D^2 - r^2) / r^2$',
'RELDIFF_LD2_LD2MAF': '% Absolute Difference in LD Scores\n$(D^2 (\\alpha = 0) - D^2 (\\alpha = 1)) / (D^2 (\\alpha = 1))$',
}
cols_to_keep = ['CHR', 'CM', 'BP', 'MAF']
frq_file = "data/genotype_files/1000G_Phase3_%s_plinkfiles/1000G.%s.QC.%d.frq"
modified_ld_score_dir = "output/ld_scores/1000G_Phase3_%s_mldscores/%s/LD.%d.l2.ldscore.gz"
pop_scores = {}
for pop in populations:
print(f">>> Reading LD Scores for {pop}...")
chr_dfs = []
frq_dfs = []
pop_outdir = os.path.join(plot_dir, pop)
makedir(pop_outdir)
pop_analysis_dir = os.path.join(analysis_dir, pop)
makedir(pop_analysis_dir)
# ----------------------------------------------------------
# *** Reading data ***
pool = Pool(num_procs)
res = pool.map(read_merge_ld_scores, list(range(22, 0, -1)))
pool.close()
pool.join()
chr_dfs = [r[0] for r in res]
frq_dfs = [r[1] for r in res]
# ----------------------------------------------------------
# *** Transforming data ***
all_snps_ld_df = pd.concat(chr_dfs, ignore_index=True)
all_snps_frq_df = pd.concat(frq_dfs)
all_snps_univar_pred = get_univariate_predictor(all_snps_ld_df, all_snps_frq_df)
eval_res = evaluate_model_differences(all_snps_univar_pred)
write_pbz2(os.path.join(pop_analysis_dir, "results.pbz2"), eval_res)
pop_scores[pop] = all_snps_univar_pred
if pop in ['EUR20', 'EUR50', 'EUR100', 'EUR200']:
continue
# ----------------------------------------------------------
# *** LD Score Comparison Plots ***
print(">> Generating scatter plots...")
plot_ld_score_comparison_scatter(all_snps_univar_pred, 'D2_1.0',
"LD Scores in 1000G (%s)" % pop,
os.path.join(pop_outdir, "scatter_R2_1.0_vs_D2_1.0_mafcol.png"),
base_score='R2_1.0',
color_col='MAF')
for y_lds in ['D2_0.0', 'D2_0.25', 'D2_0.5', 'D2_0.75']:
try:
plot_ld_score_comparison_scatter(all_snps_univar_pred, y_lds,
"LD Scores in 1000G (%s)" % pop,
os.path.join(pop_outdir, f"scatter_D2_1.0_vs_{y_lds}_mafcol.png"),
base_score='D2_1.0',
color_col='MAF')
except KeyError:
continue
"""
print(">> Generating scatter plots...")
plot_ld_score_comparison_scatter(all_snps_univar_pred, 'LD2MAF',
"LD Scores in 1000G (%s)" % pop,
os.path.join(pop_outdir, "scatter_L2MAF_vs_LD2MAF_mafcol.png"),
base_score='L2MAF',
color_col='MAF')
plot_ld_score_comparison_scatter(all_snps_univar_pred, 'LD2',
"LD Scores in 1000G (%s)" % pop,
os.path.join(pop_outdir, "scatter_LD2MAF_vs_LD2_mafcol.png"),
base_score='LD2MAF',
color_col='MAF')
plot_ld_score_comparison_scatter(all_snps_univar_pred, 'LD2',
"LD Scores in 1000G (%s)" % pop,
os.path.join(pop_outdir, "scatter_L2_vs_LD2_mafcol.png"),
color_col='MAF')
plot_ld_score_comparison_scatter(all_snps_univar_pred, 'LD2',
"LD Scores in 1000G (%s)" % pop,
os.path.join(pop_outdir, "scatter_L2_vs_LD2.png"))
plot_ld_score_comparison_scatter(all_snps_univar_pred, 'LD2MAF',
"LD Scores in 1000G (%s)" % pop,
os.path.join(pop_outdir, "scatter_L2MAF_vs_LD2MAF.png"),
base_score='L2MAF')
plot_ld_score_comparison_scatter(all_snps_univar_pred, 'LD2',
"LD Scores in 1000G (%s)" % pop,
os.path.join(pop_outdir, "scatter_LD2MAF_vs_LD2.png"),
base_score='LD2MAF')
"""
# ----------------------------------------------------------
# *** Manhattan plots ***
print(">> Generating Manhattan plots...")
"""
for lds in ld_scores:
if lds not in ['UL2', 'UL2MAF']:
plot_manhattan_ld(all_snps_univar_pred, lds,
f"LD Scores in 1000G ({pop})",
os.path.join(pop_outdir,
f"manhattan_{lds}{fig_format}"))
for lds_diff in ld_score_diff:
print(lds_diff)
_, ld1, ld2 = lds_diff.split("_")
all_snps_univar_pred[lds_diff] = all_snps_univar_pred[ld1] - all_snps_univar_pred[ld2]
#if lds_diff == 'DIFF_LD2MAF_L2MAF':
# all_snps_univar_pred.to_csv(f"./cache/{lds_diff}.csv")
plot_manhattan_ld(all_snps_univar_pred, lds_diff,
f"LD Scores in 1000G ({pop})",
os.path.join(pop_outdir, f"manhattan_{lds_diff}{fig_format}"))
for lds_diff in rel_ld_score_diff:
print(lds_diff)
_, ld1, ld2 = lds_diff.split("_")
all_snps_univar_pred[lds_diff] = 100.*np.abs(all_snps_univar_pred[ld1] -
all_snps_univar_pred[ld2]) / all_snps_univar_pred[ld2]
plot_manhattan_ld(all_snps_univar_pred, lds_diff,
f"LD Scores in 1000G ({pop})",
os.path.join(pop_outdir, f"manhattan_{lds_diff}{fig_format}"),
alpha=0.8)
plot_ld_score_comparison_scatter(all_snps_univar_pred, lds_diff,
"LD Scores in 1000G (%s)" % pop,
os.path.join(pop_outdir, f"scatter_L2MAF_vs_{lds_diff}_mafcol.png"),
base_score=ld2,
color_col='MAF')
"""
pop_scores[pop] = all_snps_univar_pred
"""
print(">>> Generating cross-population plots...")
for pop1, pop2 in [('EUR', 'AFR'), ('EUR', 'EAS'), ('EAS', 'AFR')]:
popvs_outdir = os.path.join(plot_dir, pop1 + '_vs_' + pop2)
makedir(popvs_outdir)
for lds in ld_scores + ld_score_diff:
plot_cross_population_comparison(pop_scores[pop1],
pop1,
pop_scores[pop2],
pop2,
lds,
os.path.join(popvs_outdir, lds + fig_format))
"""
print(">>> Generating sample size plots...")
plot_sample_size_comparison()